Development of personality characteristics over time

III Empirical studies

Study 2: Development of personality characteristics over time

Feichtinger, P., & Höner, O. (2015). Talented football players’ development of achievement motives, volitional compo-nents, and self-referential cognitions: A longitudinal study. This is the authors accepted manuscript of an article published as the version of record in European Journal of Sport Science 2015. http://www.tandfonline.com/

http://www.tandfonline.com/doi/full/10.1080/17461391.2015.1051134. The manuscript is used as part of this dissertation with the permission of Taylor & Francis.

Abstract

Adolescence is regarded as a key developmental phase in the course of talented football players’ careers. The present study focuses on early adolescent players’ development of achievement motives, volitional components, and self-refer-ential cognitions. Based on the multidimensional and dynamic nature of talent, the development of multifaceted person-ality characteristics is an important issue in the context of sports talent research. According to previous findings in psy-chology, personality characteristics’ development is defined by both stability and change, and the current study analyses four different types: differential stability (I), mean-level change (II), individual-level change (III), and structural stability (IV). The sample consists of 151 male players in the talent development programme of the German Football Association.

Psychological diagnostics of the personality characteristics are implemented across longitudinal sections over a time pe-riod of three seasons, from the U12 to U14 age classes. The results reveal that the personality characteristics show (I) moderate test–retest correlations over one-year intervals (.43 ≤ rtt ≤ .62), and lower coefficients for a two-year period (.26

≤ rtt ≤ .53). (II) Most of the personality characteristics’ mean values differ significantly across the age classes with small effect sizes (.01 ≤ η_G² ≤ .03). (III) Only minor individual-level changes in the football players’ development are found.

(IV) The personality characteristics’ associations within a two-factor structure do not stay invariant over time. From the results of the present study, conclusions are drawn regarding the talent identification and development process.

Keywords: Soccer, psychology, personality, stability, change

Empirical studies

Introduction

The primary aim of youth development programmes in football is to identify and develop talented players with the greatest potential to succeed at an elite level (Vaeyens et al., 2008). However, the identification and development process is difficult because talent is an extremely complex concept (Vaeyens et al., 2013). From the perspective of sports talent research, (1) multidimensional charac-teristics are required to become an elite player (Williams & Reilly, 2000) and (2) performance and its underlying characteristics change over time due to the dynamic nature of talent (Abbott & Collins, 2004).

With regard to the multidimensional understanding of talent, recent empirical research has con-sidered a wide range of physical, physiological, sociological, and psychological characteristics and examined their relationship with football performance (e.g., Figueiredo et al., 2009). These studies provide insight into the importance of different domains (e.g. motor skills and personality) for athletic success. Nevertheless, such a broad approach can only address a limited number of characteristics within the individual dimensions, although these are regarded as multifaceted constructs as well. With a few exceptions (e.g., Huijgen et al., 2014), most of this work included only one or two psychological personality characteristics. In contrast, it seems beneficial to consider multifaceted personality char-acteristics so that sports talent research can analyse the charchar-acteristics’ associations. Taking this into account, the present study exclusively focuses on personality characteristics, which have been recog-nized to play an important role in football performance (Morris, 2000).

General models of giftedness research such as the Differentiated Model of Giftedness and Talent (DMGT; Gagné, 2010) provide a valuable theoretical foundation concerning the relevance of psy-chological characteristics for football success (Mills, Butt, Maynard, & Harwood, 2012; Vaeyens et al., 2013). The DMGT considers motivation, volition, and self-awareness (i.e. self-referential cogni-tion) as major intrapersonal catalysts that facilitate or hinder the talent development process. Recent studies found empirical evidence that achievement motives (Zuber & Conzelmann, 2014), volitional

Empirical studies

components (e.g. regulation; Toering et al., 2009), and referential cognitions (e.g. self-confidence; Reilly et al., 2000) are associated with football performance level.

Based on the dynamic nature of talent, it must be taken into account that the development of talented football players primarily takes place during adolescence, a key developmental phase in which many changes occur. With regard to the lifespan theory in developmental psychology (Baltes et al., 2006), multifaceted personality characteristics may not only change over time, but they also might develop multi-directionally. In line with this, Fisher (2008, p. 127) identified a growing em-phasis on the need to take a more developmental view in sports talent research. However, so far, only a few studies have addressed this issue by analyzing the above-mentioned personality characteristics in samples of talented youth athletes. Hohmann (2009) reported that motivational and volitional char-acteristics (i.e. achievement motivation and action control) in male youth athletes (age range between 11 and 18 years) revealed test–retest correlations over two-year intervals around rtt = .50–.70. Elbe et al. (2003) and Elbe, Szymanski, et al. (2005) found that sport-specific achievement motives and vo-litional components (self-optimisation and self-impediment) only showed marginal group-level changes in young athletes aged 12–16.

Due to the lack of studies examining personality characteristics’ development in talented youth athletes, reference to research in developmental psychology seems to be beneficial. Most psycholog-ical studies in developmental research are based on trait theories (e.g. Big Five; McCrae & John, 1992). Empirical findings revealed that personality traits are relatively stable over time, but that such characteristics are also subject to change (Specht, Egloff, & Schmukle, 2011). In this regard, previous research examined different types of stability and change (e.g. De Fruyt et al., 2006; Roberts, Wood,

& Caspi, 2008), and the present study focuses on four definitions of these two concepts: Differential stability describes the degree to which inter-individual differences in personality characteristics re-main invariant over time; mean-level change refers to the extent to which the average level of a pop-ulation changes; individual-level change describes to what degree individuals vary in the amount of

Empirical studies

intra-individual change; structural stability refers to the invariance of correlational patterns among a range of personality characteristics over time.

The conceptualisation of personality characteristics’ development in terms of such different types of stability and change also contributes important insight for sports talent research. First, differential stability reflects the extent to which the relative ordering of individuals in a given characteristic changes over time, and therefore this aspect is regarded as a prerequisite for predicting behaviour (e.g. performance; Hohmann, 2009). Second, consideration of mean-level change provides relevant information on how talented athletes’ characteristics develop in general (Vaeyens et al., 2013). Third, the examination of individual-level change is of particular interest because talent research per se fo-cuses on individual differences (Ackerman, 2014). Such change analyses are important to better un-derstand the development of talented athletes in terms of variations across age classes and the poten-tial effectiveness of sport psychological interventions. Finally, if sports talent research intends to con-sider the relationship between multifaceted personality characteristics and performance, then struc-tural stability contributes insights into the complex interplay among the individual characteristics over time. Moreover, supposing that the characteristics’ associations may change, then their com-bined predictive value for athletic success may be different depending on the age group (Reilly et al., 2000). In this context, the present study focuses on the development of achievement motives, voli-tional components, and self-referential cognitions in talented football players during early adoles-cence. For this purpose, the above-mentioned four types of stability and change were analysed: dif-ferential stability (I), mean-level change (II), individual-level change (III), and structural stability (IV).

Method

Sample and design

The present study was conducted with players in the talent development programme of the Ger-man Football Association (Deutscher Fußball-Bund, DFB). The participants were among the top 4%

Empirical studies

of all eligible German players in the under 12 to under 14 (U12–U14) age classes. Psychological diagnostics of personality characteristics were implemented across longitudinal sections over a time period of three seasons, from 2010/11 to 2012/13. All of the characteristics were assessed once per season at intervals of one year. The initial study sample consisted of 828 male football players (Mage

= 11.51, SDage = 0.27 years). All of these players participated in the psychological diagnostics that were carried out in the U12 during the season 2010/11. A subsample of 151 players (Mage = 11.50, SDage = 0.27 years), who additionally attended the psychological diagnostics conducted in the 2011/12 and 2012/13 seasons, was used to analyse the personality characteristics’ development from U12 to U14.

Measures

The psychological diagnostics in the DFB talent development programme capture achievement motives, volitional components, and self-referential cognitions. To assess these characteristics, the German versions of already established self-report questionnaires were used in a football-specific and age-appropriate adaptation (items’ wording). The questionnaires were implemented as an Internet-based survey, and the individual scales demonstrated satisfactory psychometric properties in terms of reliability and validity (Feichtinger & Höner, 2014). The short scale of the “Achievement Motives Scale-Sport” by Wenhold et al. (2009a) was used to measure the two motive components: hope for success and fear of failure. The questionnaire “Volitional Components in Sport” by Wenhold et al.

(2009c) was applied to assess volitional skills (self-optimisation) and deficits (self-impediment, lack of initiation and loss of focus). Feichtinger and Höner (2014) found rather weak internal consistency of the football-specific subscale self-impediment (Cronbach’s alpha α = .64; average inter-item cor-relation rii = .17) and comparatively high correlations of the subscale lack of initiation with the re-maining scales. Therefore, the two scales were excluded from the present analyses. The “Physical Self-Concept Scales” by Stiller et al. (2004) capture the subjective perception of an athlete’s own physical abilities. The present Internet-based survey assessed the physical self-concept with regard to

Empirical studies

the motor performance tests in the DFB talent development programme (Höner et al., 2015) that measure football-specific motor skills, such as speed, agility, dribbling, ball control, and shooting (e.g. “I can sprint faster than most football players who are my age”). In addition, based on the original questionnaire’s subscale “general athleticism”, a scale assessing the general football-specific physical self-concept was included (e.g. “I play football better than most of my teammates”). The question-naire “Self-efficacy in Football” by Gerlach (2004) captures the subjective belief that one is able to perform a certain action on the basis of one’s own abilities.

Procedures

The psychological diagnostics in the DFB talent development programme were executed with the EFS Internet-based survey software 6.0−9.1. All of the participants received an informational letter that included information regarding the aim, content, and implementation of the survey as well as an Internet link and password. Players could participate at any time from any Internet-connected com-puter within a time frame of six weeks. The implementation of the psychological diagnostics was based on the former version of the Declaration of Helsinki by the World Medical Association, and the research was approved by the scientific board of the DFB and the Ethics Department of the Faculty of Economics and Social Sciences at the University of Tübingen. As part of the data privacy policy, the players were informed that participation in the survey was voluntary, all data would be stored anonymously for scientific purposes, and only employees of the DFB talent development pro-gramme’s scientific support would have access to the data. In addition, all players’ parents provided informed consent to record and use data for scientific research.

Data analysis

Statistical analyses in this study were conducted with SPSS Statistics 21 (IBM) and Mplus 5.2.1 (Muthén & Muthén). The self-report questionnaires were differently scaled; so all scales’ scores were transformed to the range [0, 1] for the purpose of comparison. The significance level was set at α = .05. As part of a preliminary drop-out analysis, a MANOVA was performed to determine whether

Empirical studies

the scales’ mean values differ between the subsample of 151 players used to examine the personality characteristics’ development (“participants”) and the remaining 677 players who were excluded from the longitudinal analyses (“drop-outs”). The multivariate mean comparison test revealed a non-sig-nificant difference between the two groups (F = 1.94; p = .06), which implies that no systematic sample selection effect occurred. This result corresponds to the multiple reasons for players’ drop-ping-out of the longitudinal diagnostics, such as deselection from the DFB talent development pro-gramme (e.g. due to coaches’ decisions), selection for a higher performance level (e.g. scouted by youth academies), or non-attendance due to personal causes (e.g. players’ decisions not to continue participating).

One-year and two-year differential stabilities (I) were analysed by calculating the scales’ test–

retest correlations for the time intervals U12–U13, U13–U14, and U12–U14. Repeated-measures ANOVAs were conducted to examine whether significant mean-level changes (II) took place across the age classes U12 to U14. Whenever the assumption of sphericity was violated, a correction of the degrees of freedom according to Greenhouse-Geisser was carried out. Generalised eta squared (Bakeman, 2005) was calculated as the effect size, with .01 ≤ η_G² < .06 as small, .06 ≤ η_G² < .14 as medium, and η_G² ≥ .14 as large effects (Cohen, 1988). A post hoc analysis computed paired sample t-tests to examine differences between the individual age classes as well as the characteristics’ change pattern in terms of increase or decrease.

Individual-level change (III) was analysed over the two-year interval from U12 to U14. For this purpose, the participants were classified as having decreased, increased, or unchanged scores by using the Reliable Change Index (RCI; Jacobson & Truax, 1991): RCI = (x2 – x1) / sdiff, where x1 and x2

represent a player’s scale scores at Time 1 and 2, respectively; sdiff is the standard error of differences between the two scores which can be computed using the standard error of measurement: 𝑠_diff =

√2(𝑠_E)², with 𝑠_E= 𝑠₁√1 − 𝑟xx. RCI scores within the interval [−1.96, 1.96] would be expected if no

Empirical studies

reliable change had occurred, whereas values outside this interval implied a reliable decrease or in-crease. In accordance with Roberts, Caspi, and Moffitt (2001), the study further examined whether significant differences occurred in the course of individuals’ development. Hence, it was analysed whether the observed distribution of RCI scores deviated from a random change pattern (i.e. 2.5%

each reliably decrease and increase, 95% remain the same) by using chi-square tests.

With regard to the personality characteristics’ associations, an EFA examining a wide range of characteristics assigned to the areas of motivation, volition, self-referential cognition, and emotion by Feichtinger and Höner (2014) revealed that achievement motives and volitional components were assigned to one common factor (“MoVo”), and the self-referential cognitions (physical self-concept and self-efficacy) were loaded on a different factor (“SeCo”). In a preliminary step of the current study, the scales’ assignment to the factors was cross-validated using a CFA with the initial sample of 828 players, and the study’s seven scales were applied. The two-factor structure (Figure 8a) showed satisfactory fit indices in the U12 (χ² = 62.83, p < .05; CFI = .98; TLI = .96; RMSEA = .07; SRMR

= .05). Based on this result, the present research examined the personality characteristics’ structural stability (IV) by computing two additional CFAs in the U13 and U14 age classes using the subsample of 151 participants. Acceptable fit indices of the two-factor structure imply invariance of the person-ality characteristics’ associations over time. For the analysis of the model fit, common conventions were used (acceptable fit indices are close to .05 for RMSEA/SRMR and close to .95 for CFI/TLI;

Schumacker & Lomax, 2010).

Results

Table 7 illustrates the results concerning the personality characteristics’ differential stabilities and mean-level changes. In general, moderate differential stabilities (I) were found for both one-year test–

retest intervals from U12–U13 (.43 ≤ rtt ≤ .58) and from U13–U14 (.44 ≤ rtt ≤ .62). The stability coefficients for the two-year period between U12 and U14 were considerably smaller (.26 ≤ rtt ≤ .53).

Table 7. Differential stability and mean-level change

Characteristics Scales

Differential stability rtt

Mean-level change M (SD)

Items α N U12-U13 U13-U14 U12-U14 U12 U13 U14 F 𝛈_𝐆^𝟐 Post-hoc

Achievement Motives

Hope for success 5 .69 151 .43 .57 .26 0.80 (0.16) 0.77 (0.17) 0.76 (0.17) 3.47* .01 U12>U14 Fear of failure 5 .72 151 .51 .44 .31 0.21 (0.15) 0.14 (0.14) 0.18 (0.15) 13.13* .03 U12>U13; U13<U14 Volitional

Components

Self-optimisation 29 .90 151 .58 .59 .53 0.82 (0.11) 0.84 (0.11) 0.81 (0.11) 6.19* .01 U13>U14

Loss of focus 9 .82 151 .53 .50 .29 0.11 (0.12) 0.09 (0.12) 0.12 (0.14) 4.65* .01 U13<U14

Physical Self-Concept

General self-concept 6 .74 151 .52 .53 .38 0.86 (0.09) 0.86 (0.09) 0.85 (0.10) 0.83

Specific self-concept 5 .83 151 .51 .62 .39 0.75 (0.15) 0.74 (0.13) 0.71 (0.13) 7.10* .02 U13>U14; U12>U14 Self-Efficacy Self-efficacy 11 .75 151 .57 .60 .48 0.88 (0.08) 0.88 (0.08) 0.87 (0.09) 2.53

Note. α = Cronbach’s alpha (taken from Feichtinger & Höner, 2014); Post-hoc = Significant mean differences (based on paired sample t-tests); * = p < .05.

Empirical studies

With regard to the individual characteristics, hope for success and fear of failure revealed, in parts, low one-year test–retest correlations, which also resulted in small two-year stabilities. Self-optimisa-tion and self-efficacy consistently showed satisfactory differential stabilities, and the other character-istics (loss of focus, general, and specific self-concept) revealed minor two-year stabilities.

In addition, significant mean-level changes (II) across the U12–U14 age classes with small effect sizes (3.47 ≤ F ≤ 13.13; p < .05; .01 ≤ η_G² ≤ .03) were found for all of the scales, except for the general self-concept (F = 0.83; p = .43) and self-efficacy (F = 2.53; p = .08). The post hoc analysis revealed that significant differences in the majority of the characteristics occurred between the U13 and U14 age classes. Additionally, the positive connoted personality characteristics’ average level tended to decrease (hope for success, self-optimisation, and specific self-concept). In contrast, both negative connoted characteristics showed a different change pattern. The volitional deficit loss of focus ten-dentially increased, and fear of failure did not show any clear trend.

Table 8 outlines the frequency distributions of the RCI values and the results of the chi-square tests with regard to the personality characteristics’ individual-level change (III) over the two-year interval from U12 to U14. No reliable change occurred for 91.39–96.03% of the participants, depend-ing on the personality characteristics. A small minority of the players showed a reliable decrease (0.66–5.96%) or increase (0.66–4.64%). Except for self-efficacy (χ² = 7.45; p < .05), all of the scales showed non-significant chi-square values (0.43 ≤ χ² ≤ 3.60; p > .05), indicating that the frequency distributions did not significantly deviate from a random change pattern. The significant deviations from the average level in self-efficacy were mainly due to the larger number of players who decreased in this particular characteristic.

Table 8. Frequency distributions of the RCI values and results of the chi-square tests

Reliable decrease RCI < -1.96

fo (%)

No reliable change -1.96 ≤ RCI ≤ 1.96

fo (%)

Reliable increase RCI > 1.96

fo (%) _df=2² Scales

Hope for success 151 6 (3.97) 142 (94.04) 3 (1.99) 1.49

Fear of failure 151 6 (3.97) 144 (95.36) 1 (0.66) 3.35

Self-optimisation 151 7 (4.64) 141 (93.38) 3 (1.99) 2.96

Loss of focus 151 2 (1.32) 142 (94.04) 7 (4.64) 3.60

General self-concept 151 4 (2.65) 142 (94.04) 5 (3.31) 0.43

Specific self-concept 151 1 (0.66) 145 (96.03) 5 (3.31) 2.45

Self-efficacy 151 9 (5.96) 138 (91.39) 4 (2.65) 7.45^*

fe 3.78 (2.50) 143.45 (95.00) 3.78 (2.50)

Note. RCI = Reliable Change Index; fo = observed frequency, fe = expected frequency; * = p < .05.

Empirical studies

The analysis of structural stability (IV; Figure 8b) revealed non-satisfactory fit indices in the U13 (χ² = 39.99, p < .05; CFI = .93; TLI = .88; RMSEA = .12; SRMR = .07) and acceptable indices in the U14 (χ² = 35.04, p < .05; CFI = .95; TLI = .92; RMSEA = .11; SRMR = .05). These results imply that the personality characteristics’ associations are not invariant over time. In line with this, the find-ings showed increasing inter-correlations between the two latent factors (rU12 = .57; rU13 = .70; rU14 = .76). Furthermore, the relevance of fear of failure increased within the factor “MoVo” (βU12 = −.59;

βU13 = −.58; βU14 = −.72), and a growing relevance of self-efficacy within the factor “SeCo” was observed (βU12 = .70; βU13 = .73; βU14 = .79).

Figure 8. CFAs (a) to cross-validate a two-factor structure (see Feichtinger & Höner, 2014) in the U12 age class and (b) to analyse the structural stability across the U13 and U14 age classes.

MoVo = Factor 1: achievement motives & volitional components; SeCo = Factor 2: self-referential cognitions; HS = Hope for success; FF = Fear of failure; SO = Self-optimisation;

LF = Loss of focus; PSC (g.) = General physical self-concept; PSC (s.) = Specific physical self-concept; SE = Self-efficacy.

Empirical studies

Discussion

The current study focused on the development of achievement motives, volitional components, and self-referential cognitions in talented football players during early adolescence. In accordance with research in developmental psychology, the study analysed four different types of stability and change. First, the personality characteristics showed moderate differential stabilities (I) over one-year test–retest intervals and lower coefficients for the two-year period. These results imply that, to a cer-tain degree, the relative ordering of individuals in such characteristics changes over time. With regard to other sport-specific research with male athletes of a similar age range, the study revealed smaller test–retest correlations (two-year correlations around .50–.70 for achievement motivation and action control; Hohmann, 2009). One possible explanation for the different level of stability could be that the present study examined a sample of all talented football players and that minor variances within such a homogenous group may have led to relatively small test–retest correlations. In line with this, Höner et al. (2015) examined the development of motor skills within the same population of football players in the DFB talent development programme. Referring to this study, the extent of the person-ality characteristics’ differential stability is comparable to one-year test–retest correlations of

Im Dokument Psychological diagnostics in the talent development program of the German Soccer Association: A stepwise procedure to examine the relevance of personality characteristics in talented soccer players (Seite 57-75)